MECHANISTIC KINETIC ANALYSIS POWERED BY ARTIFICIAL INTELLIGENCE (KINET^AI)
Lead Research Organisation:
University of Manchester
Department Name: Chemistry
Abstract
Understanding the mechanisms of catalytic organic reactions is essential for advancing the design of new catalysts, exploring novel modes of reactivity, and developing greener, more sustainable chemical processes. Mechanistic understanding provides invaluable insights, empowering chemists to enhance chemical systems and discover new reactions in a rational and informed manner. Kinetic analysis lies at the heart of mechanistic investigations, enabling researchers to test their hypotheses directly using experimental data, allowing them to discard inconsistent proposals. However, with a few recent exceptions, current kinetic analysis pipelines rely mostly on techniques developed nearly a century ago and require the derivation of complex rate law equations involving multiple mathematical approximations that limit their applicability. Furthermore, by focusing on "kinetic orders" of reagents and catalysts, these techniques often miss out on much of the rich information present in reaction kinetic profiles.
In this work, we aim to harness the power of artificial intelligence to create a more robust and comprehensive approach to kinetic analysis. Specifically, we aim to apply machine learning to the challenge of creating a model capable of processing experimental kinetic data, extracting all kinetic information, and subsequently use this information to automatically propose one or more mechanisms that are compatible with the data. We aim to develop models able to tackle various types of catalytic reactions involving one and two substrates, covering the vast majority of reactions used in academic and industrial organic chemistry laboratories.
This work will lead to the development of tools that will benefit scientists in both academia and industry streamlining the development of efficient synthetic methodologies in sectors such as pharmaceuticals, agrochemicals and others that rely on the synthesis of compounds through catalytic processes.
In this work, we aim to harness the power of artificial intelligence to create a more robust and comprehensive approach to kinetic analysis. Specifically, we aim to apply machine learning to the challenge of creating a model capable of processing experimental kinetic data, extracting all kinetic information, and subsequently use this information to automatically propose one or more mechanisms that are compatible with the data. We aim to develop models able to tackle various types of catalytic reactions involving one and two substrates, covering the vast majority of reactions used in academic and industrial organic chemistry laboratories.
This work will lead to the development of tools that will benefit scientists in both academia and industry streamlining the development of efficient synthetic methodologies in sectors such as pharmaceuticals, agrochemicals and others that rely on the synthesis of compounds through catalytic processes.
